[
https://issues.apache.org/jira/browse/HBASE-11367?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14127465#comment-14127465
]
Andrew Purtell commented on HBASE-11367:
----------------------------------------
bq. There are few changes to the ReplicationPeer and ReplicationPeers interface
though they are marked private is it ok to change the interface in 0.98.7
releases? In general what could be the best policy that could be applied here?
The best policy is minimum changes to replication code. Let's not through this
backport rewrite 0.98 replication to become branch-1/master replication. Maybe
it could be possible to introduce a pluggable replication API to 0.98 that
looks really similar to the branch-1/master API without significant code change
to replication internals in 0.98. That *could* be acceptable but replication is
a fragile area so big change probably means no. I suggest opening a backport
JIRA because making a pluggable replication endpoint in 0.98 seems a separate
subject now.
> Pluggable replication endpoint
> ------------------------------
>
> Key: HBASE-11367
> URL: https://issues.apache.org/jira/browse/HBASE-11367
> Project: HBase
> Issue Type: Sub-task
> Reporter: Enis Soztutar
> Assignee: Enis Soztutar
> Priority: Blocker
> Fix For: 0.99.0, 2.0.0
>
> Attachments: 0001-11367.patch, hbase-11367_v1.patch,
> hbase-11367_v2.patch, hbase-11367_v3.patch, hbase-11367_v4.patch,
> hbase-11367_v4.patch, hbase-11367_v5.patch
>
>
> We need a pluggable endpoint for replication for more flexibility. See parent
> jira for more context.
> ReplicationSource tails the logs for each peer. This jira introduces
> ReplicationEndpoint which is customizable per peer. ReplicationEndpoint is
> run in the same RS process and instantiated per replication peer per region
> server. Implementations of this interface handle the actual shipping of WAL
> edits to the remote cluster.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
